web/audio: Dynamic buffer size

[torokati44]

This is not done yet, just pushing for visibility.

The goal is to:

• Make demanding content (especially on mobile) as stutter-free as possible
• Reduce latency on lighter content (or on powerful hardware) as much as possible

[ousia]

@torokati44,

I wonder whether the SWF mentioned in #7532 (https://www.free-culture.tk/lfc-presentation.swf) could be a good test case.

That issue describes that playing the animation at full screen (it is actually a presentation) without hardware acceleration causes suboptimal audio playing (display is a 27″-screen).

What I also accidentally discovered some days ago is that Ruffle has problems playing sound from https://www.free-culture.tk/lfc-presentation.swf, but not from https://www.free-culture.tk/free_culture.swf.

The second link in last paragraph points to the original presentation (made with Macromedia Flash 8 [to the best of my knowledge]) and the first link points to a version made with SWFTools from the same sources.

I’m going to investigate the issue further. Many thanks for your help.

[torokati44]

Thank you for the suggestions @ousia! Once I get something that actually works, I might test those too!

[ousia]

Many thanks for your work on this PR, @torokati44.

BTW, if you feel in the mood, it would be great to have Sound.duration() info for non-standard frame rates (#6673).

[torokati44]

At least in my limited testing, this now appears to do what it's supposed to:

• z0r.de/690 no longer stutters sometimes
• z0r.de/6268 now feels more "responsive" to music

More investigation is needed though, to make sure.

[ousia]

@torokati44,

I would be more than happy to help, but I don‘t know how to get selfhosted build.

Many thanks for your excellent work.

[Dinnerbone]

Looks good to me but I'm definitely not an expert in this area :D

[torokati44]

Thanks for checking it out! And, well, neither am I! 😅

[torokati44]

Code is one thing, but it'd be great if this could get some more testing, to make sure it doesn't cause any regressions by being too opportunistic with short buffers.

@Toad06 or @n0samu, could either of you please try it out on some of your content, if you have the time of course?

[Toad06]

Forgot to mention it but I tried the PR right after it was no longer a draft.

I tested many files (including #6611 and #1673 which had issues only affecting the web build) and haven't found any regression. So this looks good to me. 😉

[torokati44]

That's great to hear, thank you! ☺️

[n0samu]

I just did a few quick tests and this seems to be working exactly as intended! In Bloons 2, which previously had very choppy audio, now the audio is delayed but mostly smooth, except in the very most demanding sections. And games that ran smoothly seem unchanged, with no additional audio delay introduced. Nicely done! 😃

[torokati44]

Thanks to you too, and also good to hear! ^^

[ousia]

@torokati44, sorry but I’m afraid that right now I don’t have access to the device that caused #7532.

I hope to have it in a not so distant future and be able to close that issue.

Many thanks for your help and your code.